Prioritization is key when looking at performance improvements. Not everything matters, and it can get easy to get overwhelmed at the potential needs.
I suggest three steps:
1) is the system basically working? Start with the break/fix issues that your Engineering team is telling you to fix first. Get the basics working. No sense polishing a turd. Leadership can ask the, "what is this activity going to fix," and be clear what it will (and won't) improve. Some IOT systems have serious functional problems to address before getting to tuning performance. And, often these issues are not even known by the organization or are inaccurately blamed on other things. With so many components and entities involved end-to-end, IOT has lots of opportunity for misreading what the real issue is.
2) Next, listen to your field and sales teams about performance. They are your ears to the field of the biggest pain points that customers are having. Ask them the top 5 areas that need improvement (which web page load-time is hurting the most, which report, which action, which delayed device action, etc.) RANK THEM. They will like that. Assign your Engineering team to work on the top 1-2 at a time, and get the improvements out there ASAP. This will improve morale because Field and Sales feels heard, and Engineering gets the reward of getting the first improvements out. This builds organizational trust. Keep working down the list of 5, but often, the top 3 bring the most bang.
3) At every opportunity, add monitoring and metrics. If you cannot measure it and see it, you cannot improve it. "Oh, well, we did at step 2, right?" No, your customers, field and sales team were your sensors and alerts, and they are the LAST RESORT. Give your team the goal to know about problems BEFORE your customers know. Beat that drum.